Published online by Cambridge University Press: 09 August 2013
This paper proposes a robot navigation scheme using wireless visual sensors deployed in an environment. Different from the conventional autonomous robot approaches, the scheme intends to relieve massive on-board information processing required by a robot to its environment so that a robot or a vehicle with less intelligence can exhibit sophisticated mobility. A three-state snake mechanism is developed for coordinating a series of sensors to form a reference path. Wireless visual sensors communicate internal forces with each other along the reference snake for dynamic adjustment, react to repulsive forces from obstacles, and activate a state change in the snake body from a flexible state to a rigid or even to a broken state due to kinematic or environmental constraints. A control snake is further proposed as a tracker of the reference path, taking into account the robot's non-holonomic constraint and limited steering power. A predictive control algorithm is developed to have an optimal velocity profile under robot dynamic constraints for the snake tracking. They together form a unified solution for robot navigation by distributed sensors to deal with the kinematic and dynamic constraints of a robot and to react to dynamic changes in advance. Simulations and experiments demonstrate the capability of a wireless sensor network to carry out low-level control activities for a vehicle.